Digital Image Processing

What is DIP?
• Digital image processing deals with
manipulation of digital images through a
digital computer.
Other applications:
• Security: biometrics

• Medical Imaging

And many more!!

• Resolution of an Image????

• 1024 x 768, 1280 x 1024, 1920 x 1080 ???

• Resolution :Number of pixels in an image.
• 2048 x 1536 : An image that is 2048 pixels
wide and 1536 pixels high.
• It contains (multiply) 3,145,728 pixels (or 3.1
Megapixels).
Information on Pixels:
Manipulation of pixel value:
• Spatial domain processing
– Direct manipulation of pixel

• Transform domain processing

– Transform the image in suitable domain
– Perform suitable operation
– Inverse transform

Which one is computationally efficient?

Spatial Domain Processing:

Intensity transformation
&
Spatial filtering
Spatial Domain processing:
• Intensity transformations:
Convert a given pixel value to a new pixel value based on some predefined function.

Which operation is this?

Some Notations:
• ‘r’ is Input image.
• ‘s’ is Output image.
• s(x,y) and r(x,y) represents arbitrary pixel value
of position (x,y) in output and input image
respectively.
• No of intensity levels present in an image is
0 toL-1.
• S(x,y)=T(r(x,y)) where T is intensity
transformation function .
• This can be represented as s=T(r) for simplicity.
Spatial filtering:
Spatial filtering:

Averaging??
Padding?
Image Enhancement:
• Image Enhancement is process of
manipulating an image so that result is more
suitable than the original to a specific
application.
Basic intensity Transformation functions:

• Used for image enhancement

• Basic types:
– Linear (identity and negative transformations)
– Logarithmic(Log and inverse log transformations)
– Power law(nth power and nth root transformations)
Identity transformation:

What is expected result??

Identity Transformation:
Negative transformation:

s = (L-1)-r

What is expected result??

Image negative:
• Used for enhancing white or gray detail
embedded in dark regions of image.
Log Transformations:
S= c log(1+r)
where c is a constant.

Narrow range of low intensity

to wider range of output.

Wider range of high intensity

to narrow range of output.

Dynamic range compression.

Power-Law Transformations:


s  cr
Where c and  are positive constant.
Gamma correction:
• A cathode ray tube (CRT), for example,
converts a video signal to light in a nonlinear
way. The light intensity I is proportional to a
power (γ) of the source voltage VS
• For a computer CRT, γ is about 1.8 to 2.5
• Viewing images properly on monitors requires
γ-correction
Piecewise –Linear Transformation:
• Contrast stretching
• Intensity level slicing
• Bit plane slicing
Contrast stretching:
• (r1,s1) and (r2,s2) controls
the shape of transformation.
• What if r1=r2,s1=0 and
s2=L-1 ??
• Its thresholding.
• What if (r1,s1)=(rmin,0)
and (r2,s2)=(rmax,L-1) ??
• Its contrast stretching.

Contrast stretching is a process that expands the range of intensity

levels in an image so that it spans the available full intensity range.
Intensity Level slicing:
• Highlighting a specific range of intensities in an
image.
• Two basic themes:
1. All the values of interest  one intensity value(say white)
All other values  another intensity value(say black).
2. All the values of interest are brighten or darken
keeping all other intensity levels unchanged.
Bit-plane slicing:
• 256 intensity levels 8 bits to represent a pixel
• It highlights the contribution made to total
image appearance by specific bits.
Bit plan slicing:
• How to obtain 8-bit plane?
• Set all intensity values from 0-127 to 0 and
128 to 255 to 1.
• How to obtain others?
• 7 bit plane: 64-127 and 192-255 1 o/w 0
• Border value =(194)d=(11000010)b
Why it is useful?
Reconstruction:
• Nth plane bit multiplied by 2^(n-1) constant.
• Addition of all such planes.
• Bit plane slicing is useful in compression and
storage
Image Histogram:
• An image histogram is a plot of the gray-level
frequencies (i.e., the number of pixels in the
image that have that gray level).
• Divide frequencies by total number of pixels (m x n image
size)to represent as probabilities.

Sum of all these probabilities?

Plot of these probabilities is probability density function of
input image pr(r)
Histogram:

x-axis – values of
intensities
y-axis – their
occurrence
Histogram:

x-axis – values of
intensities
y-axis – their
occurrence
Histogram:
• Histogram manipulation can be used for
image enhancement.
• Information inherent in histogram also is quite
useful in other image processing applications,
such as image compression and segmentation.
Histogram equalization:
• To improve the contrast of an image
• To transform an image in such a way that the
transformed image has a nearly uniform
distribution of pixel values
Histogram equalization:
Histogram Equalization:
• Intensity mapping form
s  T (r ), 0  r  L  1
Conditions:
a) T(r) is a monotonically increasing function in the
interval [0, L-1] and
b)
Histogram equalization:
Histogram Equalization:
In some formulations, we use the inverse in
which case

r  T 1 (s), 0  s  1

(a) change to
a’) T(r) is a strictly monotonically increasing
function in the interval [0, L-1]
• Let pr(r) and ps(s) denote the PDFs of r and s
respectively
dr
p s ( s )  pr ( r )
ds

• Consider following transform function:

r
s  T (r )  ( L  1)  pr ( w)dw
0

where w is a dummy variable of integration and the

right side of this equation is the cumulative
distribution function of random variable r.
Does it satisfies condition a and b?
• In discrete version:
– The probability of occurrence of gray level rk in an image is

n : the total number of pixels in the image

nk : the number of pixels that have gray level rk
L : the total number of possible gray levels in the image
– The transformation function is

– Thus, an output image is obtained by mapping each pixel with level rk

in the input image into a corresponding pixel with level sk.
r0=0s0=1 r0=0 790 s0=1
r1=1s1=3 r1=1 1023 s1=3
r2=2s2=5 r2=2 850 s2=5
r3=3s3=6 r3=3 656
s3= s4=6
r4=4s4=6 r4=4 329
r5=5s5=7 r5=5 245
r6=6s6=7 r6=6 122 s5=s6= s7=7
r7=7s7=7 r7=7 81

ps(sk)? ?
Histogram Equalization
Cumulative Histograms
 Histogram Matching (Specification)
• Histogram Specification Procedure for discrete:

Compute the histogram pr (r) of the given image

and use it to find the histogram equalization
transformation in equation
k nj
sk  T (rk )  ( L  1) , k  0,1,2,..., L  1
j 0 MN

and round the resulting values to the integer range [0, L-1]
Histogram Matching (Specification)
Compute all values of the transformation function G
using same equation
q
G( zq )  ( L  1) pz (ri ), q  0,1,2,..., L  1
i 0
and round values of G
Histogram Matching (Specification)
For every value of sk, k = 0,1,…,L-1, use the stored
values of G to find the corresponding value of zq
so that G(zq) is closest to sk and store these
mappings from s to z.

When more than one value of zq satisfies the

given sk(i.e mapping is not unique), choose
the smallest value by convention.
Example: Histogram Matching

Suppose that a 3-bit image (L=8) of size 64 × 64 pixels (MN = 4096)

has the intensity distribution shown in the following table (on the
left). Get the histogram transformation function and make the
output image with the specified histogram, listed in the table on the
right.
 Example: Histogram Matching
• Obtain the scaled histogram-equalized values,

• Compute all the values of the transformation

function G, 0
G( z0 )  7 pz ( z j )  0.00  0
j 0

G ( z1 )  0.00  0 G ( z2 )  0.00  0
G ( z3 )  1.05  1 G ( z4 )  2.45  2
G ( z5 )  4.55  5 G ( z6 )  5.95  6
G ( z7 )  7.00  7
Example: Histogram Matching

rk
0
1
2
3
4
5
6
7
Example: Histogram Matching
Example: Histogram Matching
Example: Histogram Matching
Example: Histogram Matching
Example: Histogram Matching
Histogram Matching (Specification)
• Trial and error process
• No specific rules!!!
Local Histogram processing:
• Histogram Processing methods discussed in
the previous two sections are Global.
• What if it is necessary to enhance detail over
small areas in an image?
• This procedure is to define a neighborhood
and move its center pixel to pixel.
Local Histogram processing
• At each location, the histogram of the points
in the neighborhood is computed and either a
histogram equalization or histogram
specification transformation function is
obtained.
Local Histogram processing:
• Map the intensity of the pixel centered in the
neighborhood.
• Center of the neighborhood region is then
moved to an adjacent pixel location and the
procedure is repeated
• Sometimes to reduce computation is to utilize
non overlapping regions, but this method
usually produces an undesirable “blocky”
effect.
Local Histogram processing:
Statistics:
Statistics:
• Variance of numbers is defined as:

• Alternately :
Histogram statistics:

• Alternately from sample values:

Histogram statistics:

• Alternately from sample values:

Histogram statistics:

Mean is a measure of average intensity and the variance is a

measure of contrast in an image.
Spatial filtering:
• It consist:
– Image neighborhood and sub image
– Predefined operation

The sub-image is called a filter, mask, kernel, template or

window
If operation is linear, then filter is called a linear spatial
filter otherwise nonlinear.
Basics of Spatial Filtering

Sometime we need to manipulate values obtained from neighboring

pixels

Example: How can we compute an average value of pixels in a 3x3

region center at a pixel z?

Pixel z
2 4 1 2 6 2

9 2 3 4 4 4
7 2 9 7 6 7

5 2 3 6 1 5

7 4 2 5 1 2
2 5 2 3 2 8
Image
Basics of Spatial Filtering (cont.)

Step 1: Select a pixel z.

Step 2: Select predefined neighborhood.
Step 3: select a sub image which will give u suitable result.
Step 4: Multiply elements of sub image with corresponding elements of
image neighborhood
Pixel z

2 4 1 2 6 2 1 1 1
y  3   4   4
9 2 3 4 4 4 9 9 9
1 1 1
7 2 9 7 6 7 1 1 1 1
X 1 1 1   9  7  6
5 2 3 6 1 5 9 9 9 9
1 1 1
1 1 1
7 4 2 5 1 2   3   6  1
Sub image 9 9 9
2 5 2 3 2 8
Question: How to compute the 3x3 average values at every pixels?

2 4 1 2 6 2 2 4 1 2 6 2 2 4 1 2 6 2

9 2 3 4 4 4 9 2 3 4 4 4 9 2 3 4 4 4

7 2 9 7 6 7 7 2 9 7 6 7 7 2 9 7 6 7

5 2 3 6 1 5 5 2 3 6 1 5 5 2 3 6 1 5

7 4 2 5 1 2 7 4 2 5 1 2 7 4 2 5 1 2

2 5 2 3 2 8 2 5 2 3 2 8 2 5 2 3 2 8

Solution: Imagine that we have a 3x3 window that can be placed

everywhere on the image
Spatial filtering:
• Select a pixel(x, y)
• Select neighborhood of predefined size of that
pixel.
• Select a sub image of same size as selected
neighborhood
• Multiply element of image neighborhood and
corresponding element of sub-image.
• Sum of that product: linear operation.
• Replace value of (x,y) with result.
Linear spatial filtering:
• Linear Filtering of an image f of size MxN filter
mask of size mxn is given by the expression

To generate a complete filtered image this equation must be applied for

x = 0, 1, 2, … , M-1 and y = 0, 1, 2, … , N-1
Rotation of 2D mask:
• Flip wrt one axes and then wrt other.
1 2 3 7 8 9 9 8 7
4 5 6 4 5 6 6 5 4
7 8 9 1 2 3 3 2 1
W W1

w1(-1,-1)=w(1,1), w1(-1,0)=w(1,0) ,…, w1(1,1)=w(-1,-1)

w1(s,t)=w(-s,-t) where s and t varies from -1 to 1
 3.4.2 Spatial Correlation and Convolution

• Correlation of a filter w(x,y) of size m x n with an image f(x,y)

a b
g ( x, y )    w(s, t ) f ( x  s, y  t )
s  a t  b

• Convolution of w(x,y) and f(x,y)

a b
g ( x, y )    w(s, t ) f ( x  s, y  t )
s  a t  b
Vector representation of Linear
filtering:
3.4.4 Generating Spatial Filter Masks

• Linear spatial filtering by 3 x 3 filter

9
R  w1 z1  w2 z2   w9 z9   wi zi
i 1

• Average value in 3 x 3 neighborhood

1 9
R   zi
9 i 1

• Gaussian function
x2  y 2

 e 2
2
h ( x, y )
Non linear filter :
• It requires :
– Size of neighbor hood
– Operation performed
• Operations like max, min etc
Smoothing spatial filters:
• Linear spatial filters for smoothing:
averaging filters, lowpass filters
• Noise reduction
• Undesirable side effect: blur edges

Standard average weighted

Box filter average

Which one will have less blurring??

Smoothing spatial filters:
a b
  w( s, t ) f ( x  s, y  t )
s  a t  b
g ( x, y )  a b
  w(s, t )
s  a t  b
Result of smoothing with square averaging filter masks

Original n=3

n=9
n=5

n=35
n=15
Another Smoothing Example
By smoothing the original image we get rid of
lots of the finer detail which leaves only the
gross features for thresholding

Original Image Smoothed Image Thresholded Image

15x15 mask
3.5.2 Order-Statistic (Nonlinear) Filters

• Order-statistic filters are nonlinear spatial filters

whose response is based on ordering (ranking) the pixels
• Median filter
• Replaces the pixel value by the median of the gray levels
in the neighborhood of that pixel
• Effective for impulse noise (salt-and-pepper noise)
• 3x3 neighborhood: 5th largest value
• 5x5 neighborhood: 13th largest value

• Isolated clusters of pixels that are light or dark

with respect to their neighbors, and whose area is less
than n2/2, are eliminated by an n x n median filter

• Max filter: select maximum value in the neighborhood

• Min filter: select minimum value in the neighborhood
Order statistic filters:

Original Image Averaging Mask Median filter

 Sharpening Spatial Filters
Previously we have looked at smoothing filters
which remove fine detail
Sharpening spatial filters seek to highlight fine
detail
– Remove blurring from images
– Highlight edges
– Enhances the noise
Sharpening filters are based on spatial
differentiation
 3.6.1 Foundation

• Image sharpening by first- and second-order derivatives

• Derivatives are defined in terms of differences
• Requirement of first derivative
1) Must be zero in flat areas
2) Must be nonzero at the onset (start) of step and ramp
3) Must be nonzero along ramps
• Requirement of second derivative
1) Must be zero in flat areas
2) Must be nonzero at the onset (start) of step and ramp
3) Must be zero along ramps of constant slope
 3.6.1 Foundation

f
( x)  f ( x  1)  f ( x) first-order derivative
x

f
( x  1)  f ( x)  f ( x  1)
x

2 f
 f ( x  1)  f ( x  1)  2 f ( x) second-order derivative
x 2
 3.6.1 Foundation

f
( x)  f ( x  1)  f ( x) first-order derivative
x

f
( x  1)  f ( x)  f ( x  1)
x

2 f
 f ( x  1)  f ( x  1)  2 f ( x) second-order derivative
x 2
 3.6.2 Use of Second Derivatives for Enhancement

• Isotropic filters: rotation invariant

• Simplest isotropic second-order derivative operator: Laplacian

2 f 2 f
 f  2  2
2
2-D Laplacian operation
x y

2 f
 f ( x  1, y )  f ( x  1, y )  2 f ( x, y) x-direction
x 2

2 f y-direction
 f ( x, y  1)  f ( x, y  1)  2 f ( x, y)
y 2

2 f ( x, y)  [ f ( x  1, y)  f ( x  1, y)  f ( x, y  1)  f ( x, y  1)]  4 f ( x, y)
0 1 0 1 1 1

1 -4 1 1 -8 1

0 1 0 1 1 1
Fig: a Fig: b

0 -1 0 -1 -1 -1

-1 4 -1 -1 8 -1

0 -1 0 -1 -1 -1

Fig: c Fig: d
 3.6.2 Use of Second Derivatives for Enhancement

• Image enhancement (sharpening) by Laplacian operation

 f ( x, y )   2 f ( x, y ) if the center coefficient of the

 Laplacian mask is negative
g ( x, y )  
 f ( x , y )   2
f ( x, y ) if the center coefficient of the
 Laplacian mask is positive
Unsharp Masking and Highboost Filtering :

gmask ( x, y)  f ( x, y)  f ( x, y) original image – blurred image

g ( x, y)  f ( x, y)  k  gmask ( x, y)

• When k=1, unsharp masking

• When k > 1, highboost filtering
First Derivative – The Gradient
Combining Spatial Enhancement
Methods:
Successful image enhancement is typically not
achieved using a single operation

Rather we combine a range of techniques in

order to achieve a final result
Combining Spatial Enhancement Methods

(a)
(b) (c) (d)
Laplacian filter of Sharpened version
Sobel filter of bone
bone scan (a) of bone scan
scan (a)
achieved by
subtracting (a) and
(b)
 Combining Spatial Enhancement Methods

Image (d) The product of Sharpened image

smoothed with a Result of applying a
(c) and (e) which is sum of
5*5 averaging power-law trans. to
which will be (a) and (f)
filter (g)
used as a mask
(e) (f) (g) (h)
 Combining Spatial Enhancement Methods
Compare the original and final images